What's a better stress-test program for memory than PRIME95?

[BonzaiDuck]

Well . . . . It SHOULD only work with Intel CPUs.

But I'm really wondering about the chipset. When I attempted to try TAT last year with my 680i chipset, it just popped up a message-box saying it wouldn't run.

That's my memory of it, anyway.

On another angle, this last week has been SOOOO STEWW-PID. A total waste of time, and all because -- with the resources right between my ears -- I didn't use them.

Among my sins: Failure to adjust CPU_VTT according to guidelines by the external-frequency or CPU_FSB speed. And -- get this! -- Failure to set the "Advanced Options" in PRIME95 to turn on error-checking!! THAT'S GOT to be the reason for my BSOD problem, which is now "gone." I'm guessing that with error-checking "ON," worker threads would've stopped early in the process, and there wouldn't have been BSODs. I'm also lucky to have enough experience with this motherboard to be very cautious about voltage settings. But I have some other related concerns, which I'm posting as questions in another thread.

Main conclusion here, though . . . . . my loss of confidence in PRIME95 was due to carelessness. Just . . . . plain carelessness. Even so -- it's nice to acquire OCCT as an auxiliary test program.

 

[Quiksilver]

Originally posted by: BonzaiDuck
Well . . . . It SHOULD only work with Intel CPUs.

But I'm really wondering about the chipset. When I attempted to try TAT last year with my 680i chipset, it just popped up a message-box saying it wouldn't run.

That's my memory of it, anyway.

On another angle, this last week has been SOOOO STEWW-PID. A total waste of time, and all because -- with the resources right between my ears -- I didn't use them.

Among my sins: Failure to adjust CPU_VTT according to guidelines by the external-frequency or CPU_FSB speed. And -- get this! -- Failure to set the "Advanced Options" in PRIME95 to turn on error-checking!! THAT'S GOT to be the reason for my BSOD problem, which is now "gone." I'm guessing that with error-checking "ON," worker threads would've stopped early in the process, and there wouldn't have been BSODs. I'm also lucky to have enough experience with this motherboard to be very cautious about voltage settings. But I have some other related concerns, which I'm posting as questions in another thread.

Main conclusion here, though . . . . . my loss of confidence in PRIME95 was due to carelessness. Just . . . . plain carelessness. Even so -- it's nice to acquire OCCT as an auxiliary test program.

Well I quickly scanned through the programs pages and a few users had their systems on 780i chipsets so I don't see why it wouldn't work on 680i.

 

[BonzaiDuck]

Originally posted by: Quiksilver

Originally posted by: BonzaiDuck
Well . . . . It SHOULD only work with Intel CPUs.

But I'm really wondering about the chipset. When I attempted to try TAT last year with my 680i chipset, it just popped up a message-box saying it wouldn't run.

...........
...........
Main conclusion here, though . . . . . my loss of confidence in PRIME95 was due to carelessness. Just . . . . plain carelessness. Even so -- it's nice to acquire OCCT as an auxiliary test program.

Well I quickly scanned through the programs pages and a few users had their systems on 780i chipsets so I don't see why it wouldn't work on 680i.

You mean, of course, that the 780i users were able to run INtelBurnTest / LinPack on those systems?

I guess I'll try it, then . . .

 

[Zap]

Originally posted by: error8
Why isn't this program used and advertised all around the web instead of Prime, Orthos and OCCT?

Because...

1) Not everyone knows about it.

2) Not everyone can interpret results or tune it optimally.

3) There exists... "good enough."

 

[BonzaiDuck]

NOW . . . I'm curious . . . .

 

[VirtualLarry]

You forgot the fact that linpack is actually a benchmark, not a stress-test. I'm curious as to how IntelBurnTest is doing the error-checking. Also, DuckDude, enabling error-checking in Prime95 does not avoid BSODs.

 

[Extelleron]

Originally posted by: VirtualLarry
You forgot the fact that linpack is actually a benchmark, not a stress-test. I'm curious as to how IntelBurnTest is doing the error-checking. Also, DuckDude, enabling error-checking in Prime95 does not avoid BSODs.

It is a benchmark but all you have to do is look at the results and you know whether or not you are stable. If you are unstable, the residual value will differ between the different trials. If the value remains the same, you are stable.

 

[BonzaiDuck]

Originally posted by: VirtualLarry
You forgot the fact that linpack is actually a benchmark, not a stress-test. I'm curious as to how IntelBurnTest is doing the error-checking. Also, DuckDude, enabling error-checking in Prime95 does not avoid BSODs.

Yeah! Again -- I score a "sooo STEWWW-PID" pin. I just discovered the truth of the matter you observed -- after trying to knock down the CPU_VTT setting a notch.

At this speed, it seems to need the setting I gave it -- right at the threshold of what intel considers to be the "upper limit."

 

[VirtualLarry]

"Linpack by Intel(R) is an extremely stressful program that will
put even the most powerful X86/X64 CPU in the world at its knees. Load temp
under Linpack will be up to 22*C higher than the competing software Prime95."

I've not seen that behavior. When I tried that, it didn't even hit the temps that Prime95 did. Don't ask me why.

"3. Use the same stress-testing engine that Intel uses to test their products
before they are packed and put on shelves for sale."

Sounds like creative bullshit to me. I'm willing to bet that Intel uses a test with a lot more in-depth opcode testing than just the matrix multiply stuff that linpack uses.

Edit: Do I need to download and install LinPack from Intel too, or does IntelBurnTest include those files? (Copyright violation?)

Hmm. On my laptop, I get an application crash right away. I'm using a Sempron 2800+ 32-bit cpu. Is it lacking in some SSE features that this program needs?

Edit: My chip is a Mobile Sempron 3000+ according to CPU-Z, and runs at up to 1.8Ghz, has SSE, SSE2, SSE3. No SSE4.1. Can anyone see why I might be getting a crash? Prime95 runs and is stable.

Edit: Hmm, there's a reference to "only runs on genuine Intel processors". So apparently, it crashes on purpose on AMD chips. Swell.

 

[GundamF91]

The IntelBurnTest includes the Linpack. I definitely saw more heat compared to Prime95. My 3.6Ghz under Linpack hit 75c, but in prime it's only about 68c. So it's not 22c more, but Linpack is definitely higher than Prime95.

 

[Idontcare]

Originally posted by: VirtualLarry
You forgot the fact that linpack is actually a benchmark, not a stress-test. I'm curious as to how IntelBurnTest is doing the error-checking. Also, DuckDude, enabling error-checking in Prime95 does not avoid BSODs.

Is this the program where the user must manually inspect the tabulated results from the output screen and ensure that every repeated run produced the exact same results?

What I have always wondered about these "stress test" programs is do they really test all the instructions the CPU is capable of processing?

Or do they just test a multiplication instruction in such a way that the chip generates a lot of heat which then challenges the integrity of the multiplication results?

Originally posted by: GundamF91
The IntelBurnTest includes the Linpack. I definitely saw more heat compared to Prime95. My 3.6Ghz under Linpack hit 75c, but in prime it's only about 68c. So it's not 22c more, but Linpack is definitely higher than Prime95.

Maybe Intel is referring to a 22C temp delta when using their stock HSF?

 

[BonzaiDuck]

Even if you don't think it's necessary to prove stability, it might be a good way of evaluating your DIY cooling solutions . . .

Someone else here on another thread -- I think it was Rubycon -- noted that certain media-oriented instructions appeared to be in error running some sort of video-rendering software, even if PRIME95, etc. didn't show any errs or warns.

Myself, I notice while tuning up this E8400 chip that going above 3.65 Ghz requires twisting up one of my other voltages besides the CPU_VTT -- the 1.2V_HT termination voltage for "MCP" on my nVidia chipset. Maybe that's an indication of something there. And I'm just talking over the top of my head.

I just think there's a continuum of operability over some range connected to "probability of error." And I'm pretty sure I'm at least part of the way right on the market circumstances, logical marketing and production policies for company like Intel or even AMD.

I was watching the Armand Assante remake of "On The Beach" the other night, and thinking about alpha particles and cosmic rays with their effect on RAM, it occurs to me that over-clocking or not over-clocking, all it would take to wreak complete havoc on our civilization is some "event" -- whether electromagnetic pulse, or increased radiation. Even if we survived as in "Jericho," can you imagine the nakedness people would feel after being so dependent on this technology for the time we've had it?

 

[spinejam]

The IntelBurnTest definitely heats things up more than any other program i've tested. It will expose instability much faster too!